From 95aaa207e9ef9e9f1425391826ced2ac7977fbf7 Mon Sep 17 00:00:00 2001 From: Leo Li Date: Mon, 9 Sep 2024 16:15:23 -0400 Subject: [PATCH 01/16] Revert "drm/amd/display: change the panel power savings level without a modeset" This reverts commit 8d20a066fa9beb1ec559b12945414a51d17d07e5. It's causing a failure in the abm_gradual igt test. Signed-off-by: Leo Li Signed-off-by: Alex Deucher --- .../gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c | 17 +------- drivers/gpu/drm/amd/display/dc/core/dc.c | 39 ++++++++----------- drivers/gpu/drm/amd/display/dc/dc.h | 2 - 3 files changed, 17 insertions(+), 41 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c index 1d2d5fb3c56f..6e79028c5d78 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm.c @@ -6908,14 +6908,9 @@ static ssize_t panel_power_savings_store(struct device *device, const char *buf, size_t count) { struct drm_connector *connector = dev_get_drvdata(device); - struct amdgpu_dm_connector *aconn = to_amdgpu_dm_connector(connector); struct drm_device *dev = connector->dev; - struct amdgpu_device *adev = drm_to_adev(dev); - struct dc *dc = adev->dm.dc; - struct pipe_ctx *pipe_ctx; long val; int ret; - int i; ret = kstrtol(buf, 0, &val); @@ -6930,17 +6925,7 @@ static ssize_t panel_power_savings_store(struct device *device, ABM_LEVEL_IMMEDIATE_DISABLE; drm_modeset_unlock(&dev->mode_config.connection_mutex); - mutex_lock(&adev->dm.dc_lock); - for (i = 0; i < dc->res_pool->pipe_count; i++) { - pipe_ctx = &dc->current_state->res_ctx.pipe_ctx[i]; - - if (pipe_ctx->stream && - pipe_ctx->stream->link == aconn->dc_link) { - dc_set_abm_level(dc, pipe_ctx, val); - break; - } - } - mutex_unlock(&adev->dm.dc_lock); + drm_kms_helper_hotplug_event(dev); return count; } diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index dc07ff422cde..5c39390ecbd5 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -3260,23 +3260,6 @@ fail: } -void dc_set_abm_level(struct dc *dc, struct pipe_ctx *pipe_ctx, int level) -{ - struct timing_generator *tg = pipe_ctx->stream_res.tg; - struct abm *abm = pipe_ctx->stream_res.abm; - - if (!abm) - return; - - if (tg->funcs->is_blanked && !tg->funcs->is_blanked(tg)) - tg->funcs->wait_for_state(tg, CRTC_STATE_VBLANK); - - if (level == ABM_LEVEL_IMMEDIATE_DISABLE) - dc->hwss.set_abm_immediate_disable(pipe_ctx); - else - abm->funcs->set_abm_level(abm, level); -} - static void commit_planes_do_stream_update(struct dc *dc, struct dc_stream_state *stream, struct dc_stream_update *stream_update, @@ -3405,12 +3388,22 @@ static void commit_planes_do_stream_update(struct dc *dc, dc->link_srv->set_dpms_on(dc->current_state, pipe_ctx); } - if (stream_update->abm_level) { - dc_set_abm_level(dc, pipe_ctx, - *stream_update->abm_level == - ABM_LEVEL_IMMEDIATE_DISABLE ? - ABM_LEVEL_IMMEDIATE_DISABLE : - stream->abm_level); + if (stream_update->abm_level && pipe_ctx->stream_res.abm) { + bool should_program_abm = true; + + // if otg funcs defined check if blanked before programming + if (pipe_ctx->stream_res.tg->funcs->is_blanked) + if (pipe_ctx->stream_res.tg->funcs->is_blanked(pipe_ctx->stream_res.tg)) + should_program_abm = false; + + if (should_program_abm) { + if (*stream_update->abm_level == ABM_LEVEL_IMMEDIATE_DISABLE) { + dc->hwss.set_abm_immediate_disable(pipe_ctx); + } else { + pipe_ctx->stream_res.abm->funcs->set_abm_level( + pipe_ctx->stream_res.abm, stream->abm_level); + } + } } } } diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 8c769d6519ea..3992ad73165b 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -2503,8 +2503,6 @@ void dc_z10_save_init(struct dc *dc); bool dc_is_dmub_outbox_supported(struct dc *dc); bool dc_enable_dmub_notifications(struct dc *dc); -void dc_set_abm_level(struct dc *dc, struct pipe_ctx *pipe_ctx, int level); - bool dc_abm_save_restore( struct dc *dc, struct dc_stream_state *stream, -- 2.51.0 From bc4688aed5c87ea1cc2263a1117431c50a5fd922 Mon Sep 17 00:00:00 2001 From: Lang Yu Date: Mon, 23 Sep 2024 16:59:42 +0800 Subject: [PATCH 02/16] drm/amdkfd: Remove an unused parameter in queue creation struct file *f is unused in queue creation, remove it. Signed-off-by: Lang Yu Reviewed-by: Felix Kuehling Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdkfd/kfd_chardev.c | 2 +- drivers/gpu/drm/amd/amdkfd/kfd_priv.h | 1 - drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c | 10 ++++------ 3 files changed, 5 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c index 9044bdb38cf4..a1f191a5984b 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c @@ -365,7 +365,7 @@ static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p, p->pasid, dev->id); - err = pqm_create_queue(&p->pqm, dev, filep, &q_properties, &queue_id, + err = pqm_create_queue(&p->pqm, dev, &q_properties, &queue_id, NULL, NULL, NULL, &doorbell_offset_in_process); if (err != 0) goto err_create_queue; diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h index d6530febabad..6a5bf88cc232 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h +++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h @@ -1347,7 +1347,6 @@ int pqm_init(struct process_queue_manager *pqm, struct kfd_process *p); void pqm_uninit(struct process_queue_manager *pqm); int pqm_create_queue(struct process_queue_manager *pqm, struct kfd_node *dev, - struct file *f, struct queue_properties *properties, unsigned int *qid, const struct kfd_criu_queue_priv_data *q_data, diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c index 01b960b15274..c76db22a1000 100644 --- a/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c +++ b/drivers/gpu/drm/amd/amdkfd/kfd_process_queue_manager.c @@ -235,7 +235,7 @@ void pqm_uninit(struct process_queue_manager *pqm) static int init_user_queue(struct process_queue_manager *pqm, struct kfd_node *dev, struct queue **q, struct queue_properties *q_properties, - struct file *f, unsigned int qid) + unsigned int qid) { int retval; @@ -300,7 +300,6 @@ cleanup: int pqm_create_queue(struct process_queue_manager *pqm, struct kfd_node *dev, - struct file *f, struct queue_properties *properties, unsigned int *qid, const struct kfd_criu_queue_priv_data *q_data, @@ -374,7 +373,7 @@ int pqm_create_queue(struct process_queue_manager *pqm, * allocate_sdma_queue() in create_queue() has the * corresponding check logic. */ - retval = init_user_queue(pqm, dev, &q, properties, f, *qid); + retval = init_user_queue(pqm, dev, &q, properties, *qid); if (retval != 0) goto err_create_queue; pqn->q = q; @@ -395,7 +394,7 @@ int pqm_create_queue(struct process_queue_manager *pqm, goto err_create_queue; } - retval = init_user_queue(pqm, dev, &q, properties, f, *qid); + retval = init_user_queue(pqm, dev, &q, properties, *qid); if (retval != 0) goto err_create_queue; pqn->q = q; @@ -1029,8 +1028,7 @@ int kfd_criu_restore_queue(struct kfd_process *p, print_queue_properties(&qp); - ret = pqm_create_queue(&p->pqm, pdd->dev, NULL, &qp, &queue_id, q_data, mqd, ctl_stack, - NULL); + ret = pqm_create_queue(&p->pqm, pdd->dev, &qp, &queue_id, q_data, mqd, ctl_stack, NULL); if (ret) { pr_err("Failed to create new queue err:%d\n", ret); goto exit; -- 2.51.0 From 94daae9744941ff7c355f756b192da863081f225 Mon Sep 17 00:00:00 2001 From: Jiadong Zhu Date: Fri, 20 Sep 2024 09:45:42 +0800 Subject: [PATCH 03/16] drm/amdgpu/sdma5: split out per instance resume function Extract the resume sequence from sdma_v5_0_gfx_resume for starting/restarting an individual instance. Signed-off-by: Jiadong Zhu Acked-by: Alex Deucher Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c | 253 ++++++++++++++----------- 1 file changed, 138 insertions(+), 115 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c index 3e48ea38385d..e813da1e48aa 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c @@ -705,14 +705,16 @@ static void sdma_v5_0_enable(struct amdgpu_device *adev, bool enable) } /** - * sdma_v5_0_gfx_resume - setup and start the async dma engines + * sdma_v5_0_gfx_resume_instance - start/restart a certain sdma engine * * @adev: amdgpu_device pointer + * @i: instance + * @restore: used to restore wptr when restart * - * Set up the gfx DMA ring buffers and enable them (NAVI10). - * Returns 0 for success, error for failure. + * Set up the gfx DMA ring buffers and enable them. On restart, we will restore wptr and rptr. + * Return 0 for success. */ -static int sdma_v5_0_gfx_resume(struct amdgpu_device *adev) +static int sdma_v5_0_gfx_resume_instance(struct amdgpu_device *adev, int i, bool restore) { struct amdgpu_ring *ring; u32 rb_cntl, ib_cntl; @@ -722,142 +724,163 @@ static int sdma_v5_0_gfx_resume(struct amdgpu_device *adev) u32 temp; u32 wptr_poll_cntl; u64 wptr_gpu_addr; - int i, r; - for (i = 0; i < adev->sdma.num_instances; i++) { - ring = &adev->sdma.instance[i].ring; + ring = &adev->sdma.instance[i].ring; - if (!amdgpu_sriov_vf(adev)) - WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0); + if (!amdgpu_sriov_vf(adev)) + WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_SEM_WAIT_FAIL_TIMER_CNTL), 0); - /* Set ring buffer size in dwords */ - rb_bufsz = order_base_2(ring->ring_size / 4); - rb_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL)); - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz); + /* Set ring buffer size in dwords */ + rb_bufsz = order_base_2(ring->ring_size / 4); + rb_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL)); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SIZE, rb_bufsz); #ifdef __BIG_ENDIAN - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1); - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, - RPTR_WRITEBACK_SWAP_ENABLE, 1); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_SWAP_ENABLE, 1); + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, + RPTR_WRITEBACK_SWAP_ENABLE, 1); #endif - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl); - - /* Initialize the ring buffer's read and write pointers */ + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl); + + /* Initialize the ring buffer's read and write pointers */ + if (restore) { + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), lower_32_bits(ring->wptr << 2)); + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), upper_32_bits(ring->wptr << 2)); + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), lower_32_bits(ring->wptr << 2)); + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), upper_32_bits(ring->wptr << 2)); + } else { WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR), 0); WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_HI), 0); WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), 0); WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), 0); - - /* setup the wptr shadow polling */ - wptr_gpu_addr = ring->wptr_gpu_addr; - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO), - lower_32_bits(wptr_gpu_addr)); - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI), - upper_32_bits(wptr_gpu_addr)); - wptr_poll_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, - mmSDMA0_GFX_RB_WPTR_POLL_CNTL)); - wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, - SDMA0_GFX_RB_WPTR_POLL_CNTL, - F32_POLL_ENABLE, 1); - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), - wptr_poll_cntl); - - /* set the wb address whether it's enabled or not */ - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI), - upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF); - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_LO), - lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC); - - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1); - - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE), - ring->gpu_addr >> 8); - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE_HI), - ring->gpu_addr >> 40); - + } + /* setup the wptr shadow polling */ + wptr_gpu_addr = ring->wptr_gpu_addr; + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_LO), + lower_32_bits(wptr_gpu_addr)); + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_ADDR_HI), + upper_32_bits(wptr_gpu_addr)); + wptr_poll_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, + mmSDMA0_GFX_RB_WPTR_POLL_CNTL)); + wptr_poll_cntl = REG_SET_FIELD(wptr_poll_cntl, + SDMA0_GFX_RB_WPTR_POLL_CNTL, + F32_POLL_ENABLE, 1); + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_POLL_CNTL), + wptr_poll_cntl); + + /* set the wb address whether it's enabled or not */ + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_HI), + upper_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFF); + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_RPTR_ADDR_LO), + lower_32_bits(ring->rptr_gpu_addr) & 0xFFFFFFFC); + + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RPTR_WRITEBACK_ENABLE, 1); + + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE), + ring->gpu_addr >> 8); + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_BASE_HI), + ring->gpu_addr >> 40); + + if (!restore) ring->wptr = 0; - /* before programing wptr to a less value, need set minor_ptr_update first */ - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1); + /* before programing wptr to a less value, need set minor_ptr_update first */ + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 1); - if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */ - WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), - lower_32_bits(ring->wptr << 2)); - WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), - upper_32_bits(ring->wptr << 2)); - } + if (!amdgpu_sriov_vf(adev)) { /* only bare-metal use register write for wptr */ + WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR), + lower_32_bits(ring->wptr << 2)); + WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_WPTR_HI), + upper_32_bits(ring->wptr << 2)); + } - doorbell = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL)); - doorbell_offset = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, - mmSDMA0_GFX_DOORBELL_OFFSET)); + doorbell = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL)); + doorbell_offset = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, + mmSDMA0_GFX_DOORBELL_OFFSET)); - if (ring->use_doorbell) { - doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1); - doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_GFX_DOORBELL_OFFSET, - OFFSET, ring->doorbell_index); - } else { - doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0); - } - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell); - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET), - doorbell_offset); + if (ring->use_doorbell) { + doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 1); + doorbell_offset = REG_SET_FIELD(doorbell_offset, SDMA0_GFX_DOORBELL_OFFSET, + OFFSET, ring->doorbell_index); + } else { + doorbell = REG_SET_FIELD(doorbell, SDMA0_GFX_DOORBELL, ENABLE, 0); + } + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL), doorbell); + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_DOORBELL_OFFSET), + doorbell_offset); - adev->nbio.funcs->sdma_doorbell_range(adev, i, ring->use_doorbell, - ring->doorbell_index, 20); + adev->nbio.funcs->sdma_doorbell_range(adev, i, ring->use_doorbell, + ring->doorbell_index, 20); - if (amdgpu_sriov_vf(adev)) - sdma_v5_0_ring_set_wptr(ring); + if (amdgpu_sriov_vf(adev)) + sdma_v5_0_ring_set_wptr(ring); - /* set minor_ptr_update to 0 after wptr programed */ - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0); + /* set minor_ptr_update to 0 after wptr programed */ + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_MINOR_PTR_UPDATE), 0); - if (!amdgpu_sriov_vf(adev)) { - /* set utc l1 enable flag always to 1 */ - temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL)); - temp = REG_SET_FIELD(temp, SDMA0_CNTL, UTC_L1_ENABLE, 1); - - /* enable MCBP */ - temp = REG_SET_FIELD(temp, SDMA0_CNTL, MIDCMD_PREEMPT_ENABLE, 1); - WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL), temp); - - /* Set up RESP_MODE to non-copy addresses */ - temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL)); - temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3); - temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, REDO_DELAY, 9); - WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL), temp); - - /* program default cache read and write policy */ - temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE)); - /* clean read policy and write policy bits */ - temp &= 0xFF0FFF; - temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) | (CACHE_WRITE_POLICY_L2__DEFAULT << 14)); - WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE), temp); - } + if (!amdgpu_sriov_vf(adev)) { + /* set utc l1 enable flag always to 1 */ + temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL)); + temp = REG_SET_FIELD(temp, SDMA0_CNTL, UTC_L1_ENABLE, 1); + + /* enable MCBP */ + temp = REG_SET_FIELD(temp, SDMA0_CNTL, MIDCMD_PREEMPT_ENABLE, 1); + WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_CNTL), temp); + + /* Set up RESP_MODE to non-copy addresses */ + temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL)); + temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, RESP_MODE, 3); + temp = REG_SET_FIELD(temp, SDMA0_UTCL1_CNTL, REDO_DELAY, 9); + WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_CNTL), temp); + + /* program default cache read and write policy */ + temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE)); + /* clean read policy and write policy bits */ + temp &= 0xFF0FFF; + temp |= ((CACHE_READ_POLICY_L2__DEFAULT << 12) | (CACHE_WRITE_POLICY_L2__DEFAULT << 14)); + WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_UTCL1_PAGE), temp); + } - if (!amdgpu_sriov_vf(adev)) { - /* unhalt engine */ - temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL)); - temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0); - WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), temp); - } + if (!amdgpu_sriov_vf(adev)) { + /* unhalt engine */ + temp = RREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL)); + temp = REG_SET_FIELD(temp, SDMA0_F32_CNTL, HALT, 0); + WREG32(sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_F32_CNTL), temp); + } - /* enable DMA RB */ - rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1); - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl); + /* enable DMA RB */ + rb_cntl = REG_SET_FIELD(rb_cntl, SDMA0_GFX_RB_CNTL, RB_ENABLE, 1); + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_RB_CNTL), rb_cntl); - ib_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL)); - ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1); + ib_cntl = RREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL)); + ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_ENABLE, 1); #ifdef __BIG_ENDIAN - ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1); + ib_cntl = REG_SET_FIELD(ib_cntl, SDMA0_GFX_IB_CNTL, IB_SWAP_ENABLE, 1); #endif - /* enable DMA IBs */ - WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl); + /* enable DMA IBs */ + WREG32_SOC15_IP(GC, sdma_v5_0_get_reg_offset(adev, i, mmSDMA0_GFX_IB_CNTL), ib_cntl); - if (amdgpu_sriov_vf(adev)) { /* bare-metal sequence doesn't need below to lines */ - sdma_v5_0_ctx_switch_enable(adev, true); - sdma_v5_0_enable(adev, true); - } + if (amdgpu_sriov_vf(adev)) { /* bare-metal sequence doesn't need below to lines */ + sdma_v5_0_ctx_switch_enable(adev, true); + sdma_v5_0_enable(adev, true); + } + + return amdgpu_ring_test_helper(ring); +} - r = amdgpu_ring_test_helper(ring); +/** + * sdma_v5_0_gfx_resume - setup and start the async dma engines + * + * @adev: amdgpu_device pointer + * + * Set up the gfx DMA ring buffers and enable them (NAVI10). + * Returns 0 for success, error for failure. + */ +static int sdma_v5_0_gfx_resume(struct amdgpu_device *adev) +{ + int i, r; + + for (i = 0; i < adev->sdma.num_instances; i++) { + r = sdma_v5_0_gfx_resume_instance(adev, i, false); if (r) return r; } -- 2.51.0 From fa73462dc0482644416c2a2ee042c11d93a89663 Mon Sep 17 00:00:00 2001 From: Sunil Khatri Date: Tue, 24 Sep 2024 21:30:17 +0530 Subject: [PATCH 04/16] drm/amdgpu: update the handle ptr in dump_ip_state MIME-Version: 1.0 Content-Type: text/plain; charset=utf8 Content-Transfer-Encoding: 8bit Update the ptr handle to amdgpu_ip_block ptr in all the functions. Signed-off-by: Sunil Khatri Reviewed-by: Christian König Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 +- drivers/gpu/drm/amd/amdgpu/amdgpu_job.c | 2 +- drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c | 4 ++-- drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c | 4 ++-- drivers/gpu/drm/amd/include/amd_shared.h | 4 +++- 22 files changed, 43 insertions(+), 41 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index de1f2cabdddd..1c5a4c697c7c 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -5358,7 +5358,7 @@ int amdgpu_device_pre_asic_reset(struct amdgpu_device *adev, for (i = 0; i < tmp_adev->num_ip_blocks; i++) if (tmp_adev->ip_blocks[i].version->funcs->dump_ip_state) tmp_adev->ip_blocks[i].version->funcs - ->dump_ip_state((void *)tmp_adev); + ->dump_ip_state((void *)&tmp_adev->ip_blocks[i]); dev_info(tmp_adev->dev, "Dumping IP State Completed\n"); } diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c index 16f2605ac50b..e3bb234753ef 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_job.c @@ -42,7 +42,7 @@ static void amdgpu_job_do_core_dump(struct amdgpu_device *adev, for (i = 0; i < adev->num_ip_blocks; i++) if (adev->ip_blocks[i].version->funcs->dump_ip_state) adev->ip_blocks[i].version->funcs - ->dump_ip_state((void *)adev); + ->dump_ip_state((void *)&adev->ip_blocks[i]); dev_info(adev->dev, "Dumping IP State Completed\n"); amdgpu_coredump(adev, true, false, job); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c index 45ed97038df0..61300eb88ab4 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v10_0.c @@ -9632,9 +9632,9 @@ static void gfx_v10_ip_print(void *handle, struct drm_printer *p) } } -static void gfx_v10_ip_dump(void *handle) +static void gfx_v10_ip_dump(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; uint32_t i, j, k, reg, index = 0; uint32_t reg_count = ARRAY_SIZE(gc_reg_list_10_1); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c index 33fd2da49a2a..acf2e7634200 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v11_0.c @@ -6705,9 +6705,9 @@ static void gfx_v11_ip_print(void *handle, struct drm_printer *p) } } -static void gfx_v11_ip_dump(void *handle) +static void gfx_v11_ip_dump(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; uint32_t i, j, k, reg, index = 0; uint32_t reg_count = ARRAY_SIZE(gc_reg_list_11_0); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c index 47b47d21f464..13baa6868a78 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c @@ -5102,9 +5102,9 @@ static void gfx_v12_ip_print(void *handle, struct drm_printer *p) } } -static void gfx_v12_ip_dump(void *handle) +static void gfx_v12_ip_dump(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; uint32_t i, j, k, reg, index = 0; uint32_t reg_count = ARRAY_SIZE(gc_reg_list_12_0); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c index 3a3ef93f2f43..fb4402f9bd8d 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c @@ -7377,9 +7377,9 @@ static void gfx_v9_ip_print(void *handle, struct drm_printer *p) } -static void gfx_v9_ip_dump(void *handle) +static void gfx_v9_ip_dump(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; uint32_t i, j, k, reg, index = 0; uint32_t reg_count = ARRAY_SIZE(gc_reg_list_9); diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c index 59d0c34aeacf..3d48eb281e99 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_4_3.c @@ -4645,9 +4645,9 @@ static void gfx_v9_4_3_ip_print(void *handle, struct drm_printer *p) } } -static void gfx_v9_4_3_ip_dump(void *handle) +static void gfx_v9_4_3_ip_dump(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; uint32_t i, j, k; uint32_t num_xcc, reg, num_inst; uint32_t xcc_id, xcc_offset, inst_offset; diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c index 23ef4eb36b40..50ad06f06c62 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_0.c @@ -2371,9 +2371,9 @@ static void sdma_v4_0_print_ip_state(void *handle, struct drm_printer *p) } } -static void sdma_v4_0_dump_ip_state(void *handle) +static void sdma_v4_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; uint32_t instance_offset; uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_4_0); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c index c77889040760..487cc0992263 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v4_4_2.c @@ -1878,9 +1878,9 @@ static void sdma_v4_4_2_print_ip_state(void *handle, struct drm_printer *p) } } -static void sdma_v4_4_2_dump_ip_state(void *handle) +static void sdma_v4_4_2_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; uint32_t instance_offset; uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_4_4_2); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c index e813da1e48aa..4dc8ff761888 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_0.c @@ -1822,9 +1822,9 @@ static void sdma_v5_0_print_ip_state(void *handle, struct drm_printer *p) } } -static void sdma_v5_0_dump_ip_state(void *handle) +static void sdma_v5_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; uint32_t instance_offset; uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_5_0); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c index bc9b240a3488..d19dde1d6fc5 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v5_2.c @@ -1757,9 +1757,9 @@ static void sdma_v5_2_print_ip_state(void *handle, struct drm_printer *p) } } -static void sdma_v5_2_dump_ip_state(void *handle) +static void sdma_v5_2_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; uint32_t instance_offset; uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_5_2); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c index 581fa550ef29..ed7413c1954c 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v6_0.c @@ -1624,9 +1624,9 @@ static void sdma_v6_0_print_ip_state(void *handle, struct drm_printer *p) } } -static void sdma_v6_0_dump_ip_state(void *handle) +static void sdma_v6_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; uint32_t instance_offset; uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_6_0); diff --git a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c index a8763496aed3..9e8b1ec4e6cb 100644 --- a/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c +++ b/drivers/gpu/drm/amd/amdgpu/sdma_v7_0.c @@ -1559,9 +1559,9 @@ static void sdma_v7_0_print_ip_state(void *handle, struct drm_printer *p) } } -static void sdma_v7_0_dump_ip_state(void *handle) +static void sdma_v7_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; uint32_t instance_offset; uint32_t reg_count = ARRAY_SIZE(sdma_reg_list_7_0); diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c index ecdfbfefd66a..78dfcd02d8da 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v1_0.c @@ -1957,9 +1957,9 @@ static void vcn_v1_0_print_ip_state(void *handle, struct drm_printer *p) } } -static void vcn_v1_0_dump_ip_state(void *handle) +static void vcn_v1_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; bool is_powered; uint32_t inst_off; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c index bfd067e2d2f1..5d6accdba480 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_0.c @@ -2066,9 +2066,9 @@ static void vcn_v2_0_print_ip_state(void *handle, struct drm_printer *p) } } -static void vcn_v2_0_dump_ip_state(void *handle) +static void vcn_v2_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; bool is_powered; uint32_t inst_off; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c index 04e9e806e318..8896c25fd1d4 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v2_5.c @@ -1958,9 +1958,9 @@ static void vcn_v2_5_print_ip_state(void *handle, struct drm_printer *p) } } -static void vcn_v2_5_dump_ip_state(void *handle) +static void vcn_v2_5_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; bool is_powered; uint32_t inst_off; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c index 65dd68b32280..3009f52f62d7 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v3_0.c @@ -2284,9 +2284,9 @@ static void vcn_v3_0_print_ip_state(void *handle, struct drm_printer *p) } } -static void vcn_v3_0_dump_ip_state(void *handle) +static void vcn_v3_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; bool is_powered; uint32_t inst_off; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c index 26c6f10a8c8f..17ea10c14e29 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0.c @@ -2190,9 +2190,9 @@ static void vcn_v4_0_print_ip_state(void *handle, struct drm_printer *p) } } -static void vcn_v4_0_dump_ip_state(void *handle) +static void vcn_v4_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; bool is_powered; uint32_t inst_off; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c index 0fda70336300..742bf09ae195 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_3.c @@ -1765,9 +1765,9 @@ static void vcn_v4_0_3_print_ip_state(void *handle, struct drm_printer *p) } } -static void vcn_v4_0_3_dump_ip_state(void *handle) +static void vcn_v4_0_3_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; bool is_powered; uint32_t inst_off, inst_id; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c index 9d4f5352a62c..3fb8b218e278 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v4_0_5.c @@ -1648,9 +1648,9 @@ static void vcn_v4_0_5_print_ip_state(void *handle, struct drm_printer *p) } } -static void vcn_v4_0_5_dump_ip_state(void *handle) +static void vcn_v4_0_5_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; bool is_powered; uint32_t inst_off; diff --git a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c index c305386358b4..25ae338df346 100644 --- a/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c +++ b/drivers/gpu/drm/amd/amdgpu/vcn_v5_0_0.c @@ -1375,9 +1375,9 @@ static void vcn_v5_0_print_ip_state(void *handle, struct drm_printer *p) } } -static void vcn_v5_0_dump_ip_state(void *handle) +static void vcn_v5_0_dump_ip_state(struct amdgpu_ip_block *ip_block) { - struct amdgpu_device *adev = (struct amdgpu_device *)handle; + struct amdgpu_device *adev = ip_block->adev; int i, j; bool is_powered; uint32_t inst_off; diff --git a/drivers/gpu/drm/amd/include/amd_shared.h b/drivers/gpu/drm/amd/include/amd_shared.h index 3f91926a50e9..cbb19895ddaf 100644 --- a/drivers/gpu/drm/amd/include/amd_shared.h +++ b/drivers/gpu/drm/amd/include/amd_shared.h @@ -375,6 +375,8 @@ enum amd_dpm_forced_level; * making calls to hooks from each IP block. This list is ordered to ensure * that the driver initializes the IP blocks in a safe sequence. */ +struct amdgpu_ip_block; + struct amd_ip_funcs { char *name; int (*early_init)(void *handle); @@ -399,7 +401,7 @@ struct amd_ip_funcs { int (*set_powergating_state)(void *handle, enum amd_powergating_state state); void (*get_clockgating_state)(void *handle, u64 *flags); - void (*dump_ip_state)(void *handle); + void (*dump_ip_state)(struct amdgpu_ip_block *ip_block); void (*print_ip_state)(void *handle, struct drm_printer *p); }; -- 2.51.0 From 4ae86dc8785046779db3b868a8e7ca055ddbbf8b Mon Sep 17 00:00:00 2001 From: Lijo Lazar Date: Mon, 9 Sep 2024 14:17:17 +0530 Subject: [PATCH 05/16] drm/amdgpu: Add sysfs nodes to get xcp details Add partition config nodes in sysfs to get resource instance details for a particular partition mode. A resource could be anything like an xcc, vcn decoder, system dma units etc. Details of various resource instances are available under /sys/bus/pci/devices/.../compute_partition_config/ Select a partition configuration: /sys/bus/pci/devices/.../compute_partition_config/xcp_config Number of instances of a resource: /sys/bus/pci/devices/.../compute_partition_config//num_inst Total partitions sharing the resource: /sys/bus/pci/devices/.../compute_partition_config//num_shared v2: Update node name as per spec Signed-off-by: Lijo Lazar Signed-off-by: Asad Kamal Reviewed-by: Hawking Zhang Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/amdgpu/amdgpu_device.c | 2 + drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c | 201 +++++++++++++++++++++ drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h | 5 + 3 files changed, 208 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c index 1c5a4c697c7c..f60ff041c0e2 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c @@ -4486,6 +4486,7 @@ fence_driver_init: amdgpu_fru_sysfs_init(adev); amdgpu_reg_state_sysfs_init(adev); + amdgpu_xcp_cfg_sysfs_init(adev); if (IS_ENABLED(CONFIG_PERF_EVENTS)) r = amdgpu_pmu_init(adev); @@ -4608,6 +4609,7 @@ void amdgpu_device_fini_hw(struct amdgpu_device *adev) amdgpu_fru_sysfs_fini(adev); amdgpu_reg_state_sysfs_fini(adev); + amdgpu_xcp_cfg_sysfs_fini(adev); /* disable ras feature must before hw fini */ amdgpu_ras_pre_fini(adev); diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c index a6d456ec6aeb..fc4ab1d8c7c9 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.c @@ -433,3 +433,204 @@ void amdgpu_xcp_release_sched(struct amdgpu_device *adev, } } +#define XCP_CFG_SYSFS_RES_ATTR_SHOW(_name) \ + static ssize_t amdgpu_xcp_res_sysfs_##_name##_show( \ + struct amdgpu_xcp_res_details *xcp_res, char *buf) \ + { \ + return sysfs_emit(buf, "%d\n", xcp_res->_name); \ + } + +struct amdgpu_xcp_res_sysfs_attribute { + struct attribute attr; + ssize_t (*show)(struct amdgpu_xcp_res_details *xcp_res, char *buf); +}; + +#define XCP_CFG_SYSFS_RES_ATTR(_name) \ + struct amdgpu_xcp_res_sysfs_attribute xcp_res_sysfs_attr_##_name = { \ + .attr = { .name = __stringify(_name), .mode = 0400 }, \ + .show = amdgpu_xcp_res_sysfs_##_name##_show, \ + } + +XCP_CFG_SYSFS_RES_ATTR_SHOW(num_inst) +XCP_CFG_SYSFS_RES_ATTR(num_inst); +XCP_CFG_SYSFS_RES_ATTR_SHOW(num_shared) +XCP_CFG_SYSFS_RES_ATTR(num_shared); + +#define XCP_CFG_SYSFS_RES_ATTR_PTR(_name) xcp_res_sysfs_attr_##_name.attr + +static struct attribute *xcp_cfg_res_sysfs_attrs[] = { + &XCP_CFG_SYSFS_RES_ATTR_PTR(num_inst), + &XCP_CFG_SYSFS_RES_ATTR_PTR(num_shared), NULL +}; + +ATTRIBUTE_GROUPS(xcp_cfg_res_sysfs); + +#define to_xcp_attr(x) \ + container_of(x, struct amdgpu_xcp_res_sysfs_attribute, attr) +#define to_xcp_res(x) container_of(x, struct amdgpu_xcp_res_details, kobj) + +static ssize_t xcp_cfg_res_sysfs_attr_show(struct kobject *kobj, + struct attribute *attr, char *buf) +{ + struct amdgpu_xcp_res_sysfs_attribute *attribute; + struct amdgpu_xcp_res_details *xcp_res; + + attribute = to_xcp_attr(attr); + xcp_res = to_xcp_res(kobj); + + if (!attribute->show) + return -EIO; + + return attribute->show(xcp_res, buf); +} + +static const struct sysfs_ops xcp_cfg_res_sysfs_ops = { + .show = xcp_cfg_res_sysfs_attr_show, +}; + +static const struct kobj_type xcp_cfg_res_sysfs_ktype = { + .sysfs_ops = &xcp_cfg_res_sysfs_ops, + .default_groups = xcp_cfg_res_sysfs_groups, +}; + +const char *xcp_res_names[] = { + [AMDGPU_XCP_RES_XCC] = "xcc", + [AMDGPU_XCP_RES_DMA] = "dma", + [AMDGPU_XCP_RES_DEC] = "dec", + [AMDGPU_XCP_RES_JPEG] = "jpeg", +}; + +static int amdgpu_xcp_get_res_info(struct amdgpu_xcp_mgr *xcp_mgr, + int mode, + struct amdgpu_xcp_cfg *xcp_cfg) +{ + if (xcp_mgr->funcs && xcp_mgr->funcs->get_xcp_res_info) + return xcp_mgr->funcs->get_xcp_res_info(xcp_mgr, mode, xcp_cfg); + + return -EOPNOTSUPP; +} + +#define to_xcp_cfg(x) container_of(x, struct amdgpu_xcp_cfg, kobj) +static ssize_t xcp_config_show(struct kobject *kobj, + struct kobj_attribute *attr, char *buf) +{ + struct amdgpu_xcp_cfg *xcp_cfg = to_xcp_cfg(kobj); + + return sysfs_emit(buf, "%s\n", + amdgpu_gfx_compute_mode_desc(xcp_cfg->mode)); +} + +static ssize_t xcp_config_store(struct kobject *kobj, + struct kobj_attribute *attr, + const char *buf, size_t size) +{ + struct amdgpu_xcp_cfg *xcp_cfg = to_xcp_cfg(kobj); + int mode, r; + + if (!strncasecmp("SPX", buf, strlen("SPX"))) + mode = AMDGPU_SPX_PARTITION_MODE; + else if (!strncasecmp("DPX", buf, strlen("DPX"))) + mode = AMDGPU_DPX_PARTITION_MODE; + else if (!strncasecmp("TPX", buf, strlen("TPX"))) + mode = AMDGPU_TPX_PARTITION_MODE; + else if (!strncasecmp("QPX", buf, strlen("QPX"))) + mode = AMDGPU_QPX_PARTITION_MODE; + else if (!strncasecmp("CPX", buf, strlen("CPX"))) + mode = AMDGPU_CPX_PARTITION_MODE; + else + return -EINVAL; + + r = amdgpu_xcp_get_res_info(xcp_cfg->xcp_mgr, mode, xcp_cfg); + + if (r) + return r; + + xcp_cfg->mode = mode; + return size; +} + +static struct kobj_attribute xcp_cfg_sysfs_mode = + __ATTR_RW_MODE(xcp_config, 0644); + +static void xcp_cfg_sysfs_release(struct kobject *kobj) +{ + struct amdgpu_xcp_cfg *xcp_cfg = to_xcp_cfg(kobj); + + kfree(xcp_cfg); +} + +static const struct kobj_type xcp_cfg_sysfs_ktype = { + .release = xcp_cfg_sysfs_release, + .sysfs_ops = &kobj_sysfs_ops, +}; + +void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev) +{ + struct amdgpu_xcp_res_details *xcp_res; + struct amdgpu_xcp_cfg *xcp_cfg; + int i, r, j, rid; + + if (!adev->xcp_mgr) + return; + + xcp_cfg = kzalloc(sizeof(*xcp_cfg), GFP_KERNEL); + if (!xcp_cfg) + return; + xcp_cfg->xcp_mgr = adev->xcp_mgr; + + r = kobject_init_and_add(&xcp_cfg->kobj, &xcp_cfg_sysfs_ktype, + &adev->dev->kobj, "compute_partition_config"); + if (r) + goto err1; + + r = sysfs_create_file(&xcp_cfg->kobj, &xcp_cfg_sysfs_mode.attr); + if (r) + goto err1; + + r = amdgpu_xcp_get_res_info(xcp_cfg->xcp_mgr, xcp_cfg->xcp_mgr->mode, xcp_cfg); + if (r) + goto err1; + + xcp_cfg->mode = xcp_cfg->xcp_mgr->mode; + for (i = 0; i < xcp_cfg->num_res; i++) { + xcp_res = &xcp_cfg->xcp_res[i]; + rid = xcp_res->id; + r = kobject_init_and_add(&xcp_res->kobj, + &xcp_cfg_res_sysfs_ktype, + &xcp_cfg->kobj, "%s", + xcp_res_names[rid]); + if (r) + goto err; + } + + adev->xcp_mgr->xcp_cfg = xcp_cfg; + return; +err: + for (j = 0; j < i; j++) { + xcp_res = &xcp_cfg->xcp_res[i]; + kobject_put(&xcp_res->kobj); + } + + sysfs_remove_file(&xcp_cfg->kobj, &xcp_cfg_sysfs_mode.attr); +err1: + kobject_put(&xcp_cfg->kobj); +} + +void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev) +{ + struct amdgpu_xcp_res_details *xcp_res; + struct amdgpu_xcp_cfg *xcp_cfg; + int i; + + if (!adev->xcp_mgr) + return; + + xcp_cfg = adev->xcp_mgr->xcp_cfg; + for (i = 0; i < xcp_cfg->num_res; i++) { + xcp_res = &xcp_cfg->xcp_res[i]; + kobject_put(&xcp_res->kobj); + } + + sysfs_remove_file(&xcp_cfg->kobj, &xcp_cfg_sysfs_mode.attr); + kobject_put(&xcp_cfg->kobj); +} diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h index 648237f27d1c..7ac89d78a5bf 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xcp.h @@ -68,6 +68,7 @@ struct amdgpu_xcp_res_details { enum amdgpu_xcp_res_id id; u8 num_inst; u8 num_shared; + struct kobject kobj; }; struct amdgpu_xcp_cfg { @@ -75,6 +76,7 @@ struct amdgpu_xcp_cfg { struct amdgpu_xcp_res_details xcp_res[AMDGPU_XCP_RES_MAX]; u8 num_res; struct amdgpu_xcp_mgr *xcp_mgr; + struct kobject kobj; }; struct amdgpu_xcp_ip_funcs { @@ -172,6 +174,9 @@ int amdgpu_xcp_open_device(struct amdgpu_device *adev, void amdgpu_xcp_release_sched(struct amdgpu_device *adev, struct amdgpu_ctx_entity *entity); +void amdgpu_xcp_cfg_sysfs_init(struct amdgpu_device *adev); +void amdgpu_xcp_cfg_sysfs_fini(struct amdgpu_device *adev); + #define amdgpu_xcp_select_scheds(adev, e, c, d, x, y) \ ((adev)->xcp_mgr && (adev)->xcp_mgr->funcs && \ (adev)->xcp_mgr->funcs->select_scheds ? \ -- 2.51.0 From 9dad21f910fcea2bdcff4af46159101d7f9cd8ba Mon Sep 17 00:00:00 2001 From: Yihan Zhu Date: Sat, 7 Sep 2024 13:25:19 -0400 Subject: [PATCH 06/16] drm/amd/display: update DML2 policy EnhancedPrefetchScheduleAccelerationFinal DCN35 [WHY & HOW] Mismatch in DCN35 DML2 cause bw validation failed to acquire unexpected DPP pipe to cause grey screen and system hang. Remove EnhancedPrefetchScheduleAccelerationFinal value override to match HW spec. Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Charlene Liu Signed-off-by: Yihan Zhu Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c b/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c index c4c52173ef22..11c904ae2958 100644 --- a/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c +++ b/drivers/gpu/drm/amd/display/dc/dml2/dml2_policy.c @@ -303,7 +303,6 @@ void build_unoptimized_policy_settings(enum dml_project_id project, struct dml_m if (project == dml_project_dcn35 || project == dml_project_dcn351) { policy->DCCProgrammingAssumesScanDirectionUnknownFinal = false; - policy->EnhancedPrefetchScheduleAccelerationFinal = 0; policy->AllowForPStateChangeOrStutterInVBlankFinal = dml_prefetch_support_uclk_fclk_and_stutter_if_possible; /*new*/ policy->UseOnlyMaxPrefetchModes = 1; } -- 2.51.0 From 559dc44f04ebd8d54d6493f0cd5e7d3db1fe0bb2 Mon Sep 17 00:00:00 2001 From: Samson Tam Date: Fri, 6 Sep 2024 03:28:47 -0400 Subject: [PATCH 07/16] drm/amd/display: disable adaptive scaler and sharpener when integer scaling is enabled [Why & How] When integer scaling is enabled, set taps to 1 and disable adaptive scaler and sharpener. Reviewed-by: Jun Lei Signed-off-by: Samson Tam Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc_spl_translate.c | 1 + drivers/gpu/drm/amd/display/dc/spl/dc_spl.c | 12 ++++++++++++ 2 files changed, 13 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c index 603552dbd771..20fd0afca8cb 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c +++ b/drivers/gpu/drm/amd/display/dc/dc_spl_translate.c @@ -38,6 +38,7 @@ static void populate_spltaps_from_taps(struct spl_taps *spl_scaling_quality, spl_scaling_quality->h_taps = scaling_quality->h_taps; spl_scaling_quality->v_taps_c = scaling_quality->v_taps_c; spl_scaling_quality->v_taps = scaling_quality->v_taps; + spl_scaling_quality->integer_scaling = scaling_quality->integer_scaling; } static void populate_taps_from_spltaps(struct scaling_taps *scaling_quality, const struct spl_taps *spl_scaling_quality) diff --git a/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c b/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c index 014e8a296f0c..c38a5c8646e8 100644 --- a/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c +++ b/drivers/gpu/drm/amd/display/dc/spl/dc_spl.c @@ -885,6 +885,18 @@ static bool spl_get_optimal_number_of_taps( spl_scratch->scl_data.viewport.width > max_downscale_src_width) return false; + /* Disable adaptive scaler and sharpener when integer scaling is enabled */ + if (spl_in->scaling_quality.integer_scaling) { + spl_scratch->scl_data.taps.h_taps = 1; + spl_scratch->scl_data.taps.v_taps = 1; + spl_scratch->scl_data.taps.v_taps_c = 1; + spl_scratch->scl_data.taps.h_taps_c = 1; + *enable_easf_v = false; + *enable_easf_h = false; + *enable_isharp = false; + return true; + } + /* Check if we are using EASF or not */ skip_easf = enable_easf(spl_in, spl_scratch); -- 2.51.0 From 352c3165d2b75030169e012461a16bcf97f392fc Mon Sep 17 00:00:00 2001 From: Fangzhi Zuo Date: Wed, 4 Sep 2024 15:29:24 -0400 Subject: [PATCH 08/16] drm/amd/display: Restore Optimized pbn Value if Failed to Disable DSC Existing last step of dsc policy is to restore pbn value under minimum compression when try to greedily disable dsc for a stream failed to fit in MST bw. Optimized dsc params result from optimization step is not necessarily the minimum compression, therefore it is not correct to restore the pbn under minimum compression rate. Restore the pbn under minimum compression instead of the value from optimized pbn could result in the dsc params not correct at the modeset where atomic_check failed due to not enough bw. One or more monitors connected could not light up in such case. Restore the optimized pbn value, instead of using the pbn value under minimum compression. Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Wayne Lin Signed-off-by: Fangzhi Zuo Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 17 +++++++++++++---- 1 file changed, 13 insertions(+), 4 deletions(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c index 83a31b97e96b..a08e8a0b696c 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c @@ -1027,6 +1027,7 @@ static int try_disable_dsc(struct drm_atomic_state *state, int remaining_to_try = 0; int ret; uint16_t fec_overhead_multiplier_x1000 = get_fec_overhead_multiplier(dc_link); + int var_pbn; for (i = 0; i < count; i++) { if (vars[i + k].dsc_enabled @@ -1057,13 +1058,18 @@ static int try_disable_dsc(struct drm_atomic_state *state, break; DRM_DEBUG_DRIVER("MST_DSC index #%d, try no compression\n", next_index); + var_pbn = vars[next_index].pbn; vars[next_index].pbn = kbps_to_peak_pbn(params[next_index].bw_range.stream_kbps, fec_overhead_multiplier_x1000); ret = drm_dp_atomic_find_time_slots(state, params[next_index].port->mgr, params[next_index].port, vars[next_index].pbn); - if (ret < 0) + if (ret < 0) { + DRM_DEBUG_DRIVER("%s:%d MST_DSC index #%d, failed to set pbn to the state, %d\n", + __func__, __LINE__, next_index, ret); + vars[next_index].pbn = var_pbn; return ret; + } ret = drm_dp_mst_atomic_check(state); if (ret == 0) { @@ -1071,14 +1077,17 @@ static int try_disable_dsc(struct drm_atomic_state *state, vars[next_index].dsc_enabled = false; vars[next_index].bpp_x16 = 0; } else { - DRM_DEBUG_DRIVER("MST_DSC index #%d, restore minimum compression\n", next_index); - vars[next_index].pbn = kbps_to_peak_pbn(params[next_index].bw_range.max_kbps, fec_overhead_multiplier_x1000); + DRM_DEBUG_DRIVER("MST_DSC index #%d, restore optimized pbn value\n", next_index); + vars[next_index].pbn = var_pbn; ret = drm_dp_atomic_find_time_slots(state, params[next_index].port->mgr, params[next_index].port, vars[next_index].pbn); - if (ret < 0) + if (ret < 0) { + DRM_DEBUG_DRIVER("%s:%d MST_DSC index #%d, failed to set pbn to the state, %d\n", + __func__, __LINE__, next_index, ret); return ret; + } } tried[next_index] = true; -- 2.51.0 From faee3edfcff750fe5dd760177f52978aefefeb70 Mon Sep 17 00:00:00 2001 From: Alvin Lee Date: Mon, 9 Sep 2024 16:24:05 -0400 Subject: [PATCH 09/16] drm/amd/display: Wait for all pending cleared before full update [Description] Before every full update we must wait for all pending updates to be cleared - this is particularly important for minimal transitions because if we don't wait for pending cleared, it will be as if there was no minimal transition at all. In OTG we must read 3 different status registers for pending cleared, one specifically for OTG updates, one specifically for OPTC updates, and the last for surface related updates. Reviewed-by: Dillon Varone Signed-off-by: Alvin Lee Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../drm/amd/display/dc/core/dc_hw_sequencer.c | 9 +++- .../amd/display/dc/hwss/dcn20/dcn20_hwseq.c | 4 +- .../amd/display/dc/hwss/dcn30/dcn30_hwseq.c | 27 +++++++++++ .../amd/display/dc/hwss/dcn30/dcn30_hwseq.h | 2 + .../amd/display/dc/hwss/dcn30/dcn30_init.c | 3 +- .../amd/display/dc/hwss/dcn301/dcn301_init.c | 1 + .../amd/display/dc/hwss/dcn32/dcn32_init.c | 1 + .../amd/display/dc/hwss/dcn401/dcn401_init.c | 1 + .../drm/amd/display/dc/hwss/hw_sequencer.h | 1 + .../amd/display/dc/inc/hw/timing_generator.h | 4 +- .../amd/display/dc/optc/dcn10/dcn10_optc.h | 9 ++++ .../amd/display/dc/optc/dcn20/dcn20_optc.h | 7 ++- .../amd/display/dc/optc/dcn30/dcn30_optc.c | 45 +++++++++++++++++++ .../amd/display/dc/optc/dcn30/dcn30_optc.h | 13 +++++- .../amd/display/dc/optc/dcn301/dcn301_optc.c | 3 ++ .../amd/display/dc/optc/dcn31/dcn31_optc.h | 9 +++- .../amd/display/dc/optc/dcn314/dcn314_optc.h | 9 +++- .../amd/display/dc/optc/dcn32/dcn32_optc.c | 16 ++----- .../amd/display/dc/optc/dcn32/dcn32_optc.h | 7 ++- .../amd/display/dc/optc/dcn35/dcn35_optc.h | 6 ++- .../amd/display/dc/optc/dcn401/dcn401_optc.c | 4 +- .../amd/display/dc/optc/dcn401/dcn401_optc.h | 6 ++- .../dc/resource/dcn32/dcn32_resource.h | 3 +- .../dc/resource/dcn401/dcn401_resource.h | 5 ++- 24 files changed, 161 insertions(+), 34 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c index 7ee2be8f82c4..2cb9253c9bde 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc_hw_sequencer.c @@ -1071,8 +1071,13 @@ void hwss_wait_for_outstanding_hw_updates(struct dc *dc, struct dc_state *dc_con if (!pipe_ctx->stream) continue; - if (pipe_ctx->stream_res.tg->funcs->wait_drr_doublebuffer_pending_clear) - pipe_ctx->stream_res.tg->funcs->wait_drr_doublebuffer_pending_clear(pipe_ctx->stream_res.tg); + /* For full update we must wait for all double buffer updates, not just DRR updates. This + * is particularly important for minimal transitions. Only check for OTG_MASTER pipes, + * as non-OTG Master pipes share the same OTG as + */ + if (resource_is_pipe_type(pipe_ctx, OTG_MASTER) && dc->hwss.wait_for_all_pending_updates) { + dc->hwss.wait_for_all_pending_updates(pipe_ctx); + } hubp = pipe_ctx->plane_res.hubp; if (!hubp) diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c index a80c08582932..b383ed8cb4d4 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn20/dcn20_hwseq.c @@ -2255,9 +2255,9 @@ void dcn20_post_unlock_program_front_end( struct timing_generator *tg = pipe->stream_res.tg; - if (tg->funcs->get_double_buffer_pending) { + if (tg->funcs->get_optc_double_buffer_pending) { for (j = 0; j < TIMEOUT_FOR_PIPE_ENABLE_US / polling_interval_us - && tg->funcs->get_double_buffer_pending(tg); j++) + && tg->funcs->get_optc_double_buffer_pending(tg); j++) udelay(polling_interval_us); } } diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c index bded33575493..bf9b5daea2af 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.c @@ -1185,3 +1185,30 @@ void dcn30_prepare_bandwidth(struct dc *dc, if (!dc->clk_mgr->clks.fw_based_mclk_switching) dc_dmub_srv_p_state_delegate(dc, false, context); } + +void dcn30_wait_for_all_pending_updates(const struct pipe_ctx *pipe_ctx) +{ + struct timing_generator *tg = pipe_ctx->stream_res.tg; + bool pending_updates = false; + unsigned int i; + + if (tg && tg->funcs->is_tg_enabled(tg)) { + // Poll for 100ms maximum + for (i = 0; i < 100000; i++) { + pending_updates = false; + if (tg->funcs->get_optc_double_buffer_pending) + pending_updates |= tg->funcs->get_optc_double_buffer_pending(tg); + + if (tg->funcs->get_otg_double_buffer_pending) + pending_updates |= tg->funcs->get_otg_double_buffer_pending(tg); + + if (tg->funcs->get_pipe_update_pending && pipe_ctx->plane_state) + pending_updates |= tg->funcs->get_pipe_update_pending(tg); + + if (!pending_updates) + break; + + udelay(1); + } + } +} diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.h b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.h index 6a153e7ce910..4b90b781c4f2 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_hwseq.h @@ -96,4 +96,6 @@ void dcn30_set_hubp_blank(const struct dc *dc, void dcn30_prepare_bandwidth(struct dc *dc, struct dc_state *context); +void dcn30_wait_for_all_pending_updates(const struct pipe_ctx *pipe_ctx); + #endif /* __DC_HWSS_DCN30_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_init.c index 2a8dc40d2847..0e8d32e3dbae 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn30/dcn30_init.c @@ -108,7 +108,8 @@ static const struct hw_sequencer_funcs dcn30_funcs = { .set_disp_pattern_generator = dcn30_set_disp_pattern_generator, .get_dcc_en_bits = dcn10_get_dcc_en_bits, .update_visual_confirm_color = dcn10_update_visual_confirm_color, - .is_abm_supported = dcn21_is_abm_supported + .is_abm_supported = dcn21_is_abm_supported, + .wait_for_all_pending_updates = dcn30_wait_for_all_pending_updates, }; static const struct hwseq_private_funcs dcn30_private_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn301/dcn301_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn301/dcn301_init.c index 93e49d87a67c..780ce4c064aa 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn301/dcn301_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn301/dcn301_init.c @@ -107,6 +107,7 @@ static const struct hw_sequencer_funcs dcn301_funcs = { .optimize_pwr_state = dcn21_optimize_pwr_state, .exit_optimized_pwr_state = dcn21_exit_optimized_pwr_state, .update_visual_confirm_color = dcn10_update_visual_confirm_color, + .wait_for_all_pending_updates = dcn30_wait_for_all_pending_updates, }; static const struct hwseq_private_funcs dcn301_private_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_init.c index 3422b564ae98..8e0946fd5b7f 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn32/dcn32_init.c @@ -121,6 +121,7 @@ static const struct hw_sequencer_funcs dcn32_funcs = { .is_pipe_topology_transition_seamless = dcn32_is_pipe_topology_transition_seamless, .calculate_pix_rate_divider = dcn32_calculate_pix_rate_divider, .program_outstanding_updates = dcn32_program_outstanding_updates, + .wait_for_all_pending_updates = dcn30_wait_for_all_pending_updates, }; static const struct hwseq_private_funcs dcn32_private_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c index a2ca07235c83..73a632b5ff89 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_init.c @@ -100,6 +100,7 @@ static const struct hw_sequencer_funcs dcn401_funcs = { .fams2_update_config = dcn401_fams2_update_config, .fams2_global_control_lock_fast = dcn401_fams2_global_control_lock_fast, .program_outstanding_updates = dcn401_program_outstanding_updates, + .wait_for_all_pending_updates = dcn30_wait_for_all_pending_updates, }; static const struct hwseq_private_funcs dcn401_private_funcs = { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h index ac9205625623..b8c47e4c51c1 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h +++ b/drivers/gpu/drm/amd/display/dc/hwss/hw_sequencer.h @@ -462,6 +462,7 @@ struct hw_sequencer_funcs { void (*program_outstanding_updates)(struct dc *dc, struct dc_state *context); void (*setup_hpo_hw_control)(const struct dce_hwseq *hws, bool enable); + void (*wait_for_all_pending_updates)(const struct pipe_ctx *pipe_ctx); }; void color_space_to_black_color( diff --git a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h index 3d4c8bd42b49..4e08e80eafe8 100644 --- a/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h +++ b/drivers/gpu/drm/amd/display/dc/inc/hw/timing_generator.h @@ -342,7 +342,9 @@ struct timing_generator_funcs { void (*wait_drr_doublebuffer_pending_clear)(struct timing_generator *tg); void (*set_long_vtotal)(struct timing_generator *optc, const struct long_vtotal_params *params); void (*wait_odm_doublebuffer_pending_clear)(struct timing_generator *tg); - bool (*get_double_buffer_pending)(struct timing_generator *tg); + bool (*get_optc_double_buffer_pending)(struct timing_generator *tg); + bool (*get_otg_double_buffer_pending)(struct timing_generator *tg); + bool (*get_pipe_update_pending)(struct timing_generator *tg); }; #endif diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h index b7a57f98553d..40757f20d73f 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn10/dcn10_optc.h @@ -202,6 +202,7 @@ struct dcn_optc_registers { uint32_t OPTC_CLOCK_CONTROL; uint32_t OPTC_WIDTH_CONTROL2; uint32_t OTG_PSTATE_REGISTER; + uint32_t OTG_PIPE_UPDATE_STATUS; }; #define TG_COMMON_MASK_SH_LIST_DCN(mask_sh)\ @@ -566,6 +567,12 @@ struct dcn_optc_registers { type OTG_H_TIMING_DIV_MODE_DB_UPDATE_PENDING;\ type OPTC_DOUBLE_BUFFER_PENDING;\ +#define TG_REG_FIELD_LIST_DCN2_0(type) \ + type OTG_FLIP_PENDING;\ + type OTG_DC_REG_UPDATE_PENDING;\ + type OTG_CURSOR_UPDATE_PENDING;\ + type OTG_VUPDATE_KEEPOUT_STATUS;\ + #define TG_REG_FIELD_LIST_DCN3_2(type) \ type OTG_H_TIMING_DIV_MODE_MANUAL; @@ -600,6 +607,7 @@ struct dcn_optc_registers { struct dcn_optc_shift { TG_REG_FIELD_LIST(uint8_t) + TG_REG_FIELD_LIST_DCN2_0(uint8_t) TG_REG_FIELD_LIST_DCN3_2(uint8_t) TG_REG_FIELD_LIST_DCN3_5(uint8_t) TG_REG_FIELD_LIST_DCN401(uint8_t) @@ -607,6 +615,7 @@ struct dcn_optc_shift { struct dcn_optc_mask { TG_REG_FIELD_LIST(uint32_t) + TG_REG_FIELD_LIST_DCN2_0(uint32_t) TG_REG_FIELD_LIST_DCN3_2(uint32_t) TG_REG_FIELD_LIST_DCN3_5(uint32_t) TG_REG_FIELD_LIST_DCN401(uint32_t) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.h index 364034b19028..928e110b95fb 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn20/dcn20_optc.h @@ -43,7 +43,8 @@ SRI(OPTC_MEMORY_CONFIG, ODM, inst),\ SR(DWB_SOURCE_SELECT),\ SRI(OTG_MANUAL_FLOW_CONTROL, OTG, inst), \ - SRI(OTG_DRR_CONTROL, OTG, inst) + SRI(OTG_DRR_CONTROL, OTG, inst),\ + SRI(OTG_PIPE_UPDATE_STATUS, OTG, inst) #define TG_COMMON_MASK_SH_LIST_DCN2_0(mask_sh)\ TG_COMMON_MASK_SH_LIST_DCN(mask_sh),\ @@ -53,6 +54,10 @@ SF(OTG0_OTG_GLOBAL_CONTROL2, GLOBAL_UPDATE_LOCK_EN, mask_sh),\ SF(OTG0_OTG_GLOBAL_CONTROL2, DIG_UPDATE_LOCATION, mask_sh),\ SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_RANGE_TIMING_DBUF_UPDATE_MODE, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh),\ SF(OTG0_OTG_GSL_WINDOW_X, OTG_GSL_WINDOW_START_X, mask_sh),\ SF(OTG0_OTG_GSL_WINDOW_X, OTG_GSL_WINDOW_END_X, mask_sh), \ SF(OTG0_OTG_GSL_WINDOW_Y, OTG_GSL_WINDOW_START_Y, mask_sh),\ diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c index abcd03d78668..4c95c0958612 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.c @@ -271,6 +271,48 @@ void optc3_set_odm_combine(struct timing_generator *optc, int *opp_id, int opp_c optc1->opp_count = opp_cnt; } +/* OTG status register that indicates OPTC update is pending */ +bool optc3_get_optc_double_buffer_pending(struct timing_generator *optc) +{ + struct optc *optc1 = DCN10TG_FROM_TG(optc); + uint32_t update_pending = 0; + + REG_GET(OPTC_INPUT_GLOBAL_CONTROL, + OPTC_DOUBLE_BUFFER_PENDING, + &update_pending); + + return (update_pending == 1); +} + +/* OTG status register that indicates OTG update is pending */ +bool optc3_get_otg_update_pending(struct timing_generator *optc) +{ + struct optc *optc1 = DCN10TG_FROM_TG(optc); + uint32_t update_pending = 0; + + REG_GET(OTG_DOUBLE_BUFFER_CONTROL, + OTG_UPDATE_PENDING, + &update_pending); + + return (update_pending == 1); +} + +/* OTG status register that indicates surface update is pending */ +bool optc3_get_pipe_update_pending(struct timing_generator *optc) +{ + struct optc *optc1 = DCN10TG_FROM_TG(optc); + uint32_t flip_pending = 0; + uint32_t dc_update_pending = 0; + + REG_GET_2(OTG_PIPE_UPDATE_STATUS, + OTG_FLIP_PENDING, + &flip_pending, + OTG_DC_REG_UPDATE_PENDING, + &dc_update_pending); + + return (flip_pending == 1 || dc_update_pending == 1); +} + /** * optc3_set_timing_double_buffer() - DRR double buffering control * @@ -375,6 +417,9 @@ static struct timing_generator_funcs dcn30_tg_funcs = { .get_hw_timing = optc1_get_hw_timing, .wait_drr_doublebuffer_pending_clear = optc3_wait_drr_doublebuffer_pending_clear, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, + .get_optc_double_buffer_pending = optc3_get_optc_double_buffer_pending, + .get_otg_double_buffer_pending = optc3_get_otg_update_pending, + .get_pipe_update_pending = optc3_get_pipe_update_pending, }; void dcn30_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.h index bda974d432ea..e2303f9eaf13 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn30/dcn30_optc.h @@ -109,7 +109,8 @@ SRI(OPTC_BYTES_PER_PIXEL, ODM, inst),\ SRI(OPTC_WIDTH_CONTROL, ODM, inst),\ SRI(OPTC_MEMORY_CONFIG, ODM, inst),\ - SR(DWB_SOURCE_SELECT) + SR(DWB_SOURCE_SELECT),\ + SRI(OTG_PIPE_UPDATE_STATUS, OTG, inst) #define DCN30_VTOTAL_REGS_SF(mask_sh) @@ -209,6 +210,7 @@ SF(ODM0_OPTC_INPUT_CLOCK_CONTROL, OPTC_INPUT_CLK_GATE_DIS, mask_sh),\ SF(ODM0_OPTC_INPUT_GLOBAL_CONTROL, OPTC_UNDERFLOW_OCCURRED_STATUS, mask_sh),\ SF(ODM0_OPTC_INPUT_GLOBAL_CONTROL, OPTC_UNDERFLOW_CLEAR, mask_sh),\ + SF(ODM0_OPTC_INPUT_GLOBAL_CONTROL, OPTC_DOUBLE_BUFFER_PENDING, mask_sh),\ SF(VTG0_CONTROL, VTG0_ENABLE, mask_sh),\ SF(VTG0_CONTROL, VTG0_FP2, mask_sh),\ SF(VTG0_CONTROL, VTG0_VCOUNT_INIT, mask_sh),\ @@ -319,7 +321,11 @@ SF(OTG0_OTG_DRR_V_TOTAL_CHANGE, OTG_DRR_V_TOTAL_CHANGE_LIMIT, mask_sh),\ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE, mask_sh),\ SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_PENDING, mask_sh),\ - SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_MODE, mask_sh) + SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_MODE, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh),\ void dcn30_timing_generator_init(struct optc *optc1); @@ -356,4 +362,7 @@ void optc3_set_odm_combine(struct timing_generator *optc, int *opp_id, int opp_c void optc3_wait_drr_doublebuffer_pending_clear(struct timing_generator *optc); void optc3_tg_init(struct timing_generator *optc); void optc3_set_vtotal_min_max(struct timing_generator *optc, int vtotal_min, int vtotal_max); +bool optc3_get_optc_double_buffer_pending(struct timing_generator *optc); +bool optc3_get_otg_update_pending(struct timing_generator *optc); +bool optc3_get_pipe_update_pending(struct timing_generator *optc); #endif /* __DC_OPTC_DCN30_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c index 1a22ae89fb55..d7a45ef2d01b 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn301/dcn301_optc.c @@ -169,6 +169,9 @@ static struct timing_generator_funcs dcn30_tg_funcs = { .get_hw_timing = optc1_get_hw_timing, .wait_drr_doublebuffer_pending_clear = optc3_wait_drr_doublebuffer_pending_clear, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, + .get_optc_double_buffer_pending = optc3_get_optc_double_buffer_pending, + .get_otg_double_buffer_pending = optc3_get_otg_update_pending, + .get_pipe_update_pending = optc3_get_pipe_update_pending, }; void dcn301_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h index 30b81a448ce2..fbbe86d00c2e 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn31/dcn31_optc.h @@ -99,7 +99,8 @@ SRI(OPTC_MEMORY_CONFIG, ODM, inst),\ SRI(OTG_CRC_CNTL2, OTG, inst),\ SR(DWB_SOURCE_SELECT),\ - SRI(OTG_DRR_CONTROL, OTG, inst) + SRI(OTG_DRR_CONTROL, OTG, inst),\ + SRI(OTG_PIPE_UPDATE_STATUS, OTG, inst) #define OPTC_COMMON_MASK_SH_LIST_DCN3_1(mask_sh)\ SF(OTG0_OTG_VSTARTUP_PARAM, VSTARTUP_START, mask_sh),\ @@ -254,7 +255,11 @@ SF(OTG0_OTG_CRC_CNTL2, OTG_CRC_DATA_STREAM_COMBINE_MODE, mask_sh),\ SF(OTG0_OTG_CRC_CNTL2, OTG_CRC_DATA_STREAM_SPLIT_MODE, mask_sh),\ SF(OTG0_OTG_CRC_CNTL2, OTG_CRC_DATA_FORMAT, mask_sh),\ - SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh) + SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh),\ void dcn31_timing_generator_init(struct optc *optc1); diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.h index 99c098e76116..0ff72b97b465 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn314/dcn314_optc.h @@ -98,7 +98,8 @@ SRI(OPTC_BYTES_PER_PIXEL, ODM, inst),\ SRI(OPTC_WIDTH_CONTROL, ODM, inst),\ SRI(OPTC_MEMORY_CONFIG, ODM, inst),\ - SRI(OTG_DRR_CONTROL, OTG, inst) + SRI(OTG_DRR_CONTROL, OTG, inst),\ + SRI(OTG_PIPE_UPDATE_STATUS, OTG, inst) #define OPTC_COMMON_MASK_SH_LIST_DCN3_14(mask_sh)\ SF(OTG0_OTG_VSTARTUP_PARAM, VSTARTUP_START, mask_sh),\ @@ -248,7 +249,11 @@ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE, mask_sh),\ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE_MANUAL, mask_sh),\ SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_MODE, mask_sh),\ - SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh) + SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh),\ void dcn314_timing_generator_init(struct optc *optc1); diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c index 00094f0e8470..c217f653b3c8 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.c @@ -297,18 +297,6 @@ static void optc32_set_drr( optc32_setup_manual_trigger(optc); } -bool optc32_get_double_buffer_pending(struct timing_generator *optc) -{ - struct optc *optc1 = DCN10TG_FROM_TG(optc); - uint32_t update_pending = 0; - - REG_GET(OPTC_INPUT_GLOBAL_CONTROL, - OPTC_DOUBLE_BUFFER_PENDING, - &update_pending); - - return (update_pending == 1); -} - static struct timing_generator_funcs dcn32_tg_funcs = { .validate_timing = optc1_validate_timing, .program_timing = optc1_program_timing, @@ -373,7 +361,9 @@ static struct timing_generator_funcs dcn32_tg_funcs = { .setup_manual_trigger = optc2_setup_manual_trigger, .get_hw_timing = optc1_get_hw_timing, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, - .get_double_buffer_pending = optc32_get_double_buffer_pending, + .get_optc_double_buffer_pending = optc3_get_optc_double_buffer_pending, + .get_otg_double_buffer_pending = optc3_get_otg_update_pending, + .get_pipe_update_pending = optc3_get_pipe_update_pending, }; void dcn32_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.h index 665d7c52f67c..0b0964a9da74 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn32/dcn32_optc.h @@ -177,7 +177,11 @@ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE, mask_sh),\ SF(OTG0_OTG_H_TIMING_CNTL, OTG_H_TIMING_DIV_MODE_MANUAL, mask_sh),\ SF(OTG0_OTG_DOUBLE_BUFFER_CONTROL, OTG_DRR_TIMING_DBUF_UPDATE_MODE, mask_sh),\ - SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh) + SF(OTG0_OTG_DRR_CONTROL, OTG_V_TOTAL_LAST_USED_BY_DRR, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh) void dcn32_timing_generator_init(struct optc *optc1); void optc32_set_h_timing_div_manual_mode(struct timing_generator *optc, bool manual_mode); @@ -185,6 +189,5 @@ void optc32_get_odm_combine_segments(struct timing_generator *tg, int *odm_combi void optc32_set_odm_bypass(struct timing_generator *optc, const struct dc_crtc_timing *dc_crtc_timing); void optc32_wait_odm_doublebuffer_pending_clear(struct timing_generator *tg); -bool optc32_get_double_buffer_pending(struct timing_generator *optc); #endif /* __DC_OPTC_DCN32_H__ */ diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.h index d077e2392379..be749ab41dce 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn35/dcn35_optc.h @@ -67,7 +67,11 @@ SF(OTG0_OTG_CRC1_WINDOWB_Y_CONTROL_READBACK, OTG_CRC1_WINDOWB_Y_END_READBACK, mask_sh),\ SF(OPTC_CLOCK_CONTROL, OPTC_FGCG_REP_DIS, mask_sh),\ SF(OTG0_OTG_V_COUNT_STOP_CONTROL, OTG_V_COUNT_STOP, mask_sh),\ - SF(OTG0_OTG_V_COUNT_STOP_CONTROL2, OTG_V_COUNT_STOP_TIMER, mask_sh) + SF(OTG0_OTG_V_COUNT_STOP_CONTROL2, OTG_V_COUNT_STOP_TIMER, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh) void dcn35_timing_generator_init(struct optc *optc1); diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c index a5d6a7dca554..db670fc17264 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.c @@ -493,7 +493,9 @@ static struct timing_generator_funcs dcn401_tg_funcs = { .setup_manual_trigger = optc2_setup_manual_trigger, .get_hw_timing = optc1_get_hw_timing, .is_two_pixels_per_container = optc1_is_two_pixels_per_container, - .get_double_buffer_pending = optc32_get_double_buffer_pending, + .get_optc_double_buffer_pending = optc3_get_optc_double_buffer_pending, + .get_otg_double_buffer_pending = optc3_get_otg_update_pending, + .get_pipe_update_pending = optc3_get_pipe_update_pending, }; void dcn401_timing_generator_init(struct optc *optc1) diff --git a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.h b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.h index bb13a645802d..1be89571986f 100644 --- a/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.h +++ b/drivers/gpu/drm/amd/display/dc/optc/dcn401/dcn401_optc.h @@ -159,7 +159,11 @@ SF(OTG0_OTG_PSTATE_REGISTER, OTG_PSTATE_KEEPOUT_START, mask_sh),\ SF(OTG0_OTG_PSTATE_REGISTER, OTG_PSTATE_EXTEND, mask_sh),\ SF(OTG0_OTG_PSTATE_REGISTER, OTG_UNBLANK, mask_sh),\ - SF(OTG0_OTG_PSTATE_REGISTER, OTG_PSTATE_ALLOW_WIDTH_MIN, mask_sh) + SF(OTG0_OTG_PSTATE_REGISTER, OTG_PSTATE_ALLOW_WIDTH_MIN, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_FLIP_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_DC_REG_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_CURSOR_UPDATE_PENDING, mask_sh),\ + SF(OTG0_OTG_PIPE_UPDATE_STATUS, OTG_VUPDATE_KEEPOUT_STATUS, mask_sh) void dcn401_timing_generator_init(struct optc *optc1); diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h index 7901792afb7b..86c6e5e8c42e 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn32/dcn32_resource.h @@ -1054,7 +1054,8 @@ unsigned int dcn32_calculate_mall_ways_from_bytes(const struct dc *dc, unsigned SRI_ARR(OPTC_BYTES_PER_PIXEL, ODM, inst), \ SRI_ARR(OPTC_WIDTH_CONTROL, ODM, inst), \ SRI_ARR(OPTC_MEMORY_CONFIG, ODM, inst), \ - SRI_ARR(OTG_DRR_CONTROL, OTG, inst) + SRI_ARR(OTG_DRR_CONTROL, OTG, inst), \ + SRI_ARR(OTG_PIPE_UPDATE_STATUS, OTG, inst) /* HUBP */ diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h index 514d1ce20df9..bdafa7496cea 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.h @@ -536,8 +536,9 @@ void dcn401_prepare_mcache_programming(struct dc *dc, struct dc_state *context); SRI_ARR(OPTC_WIDTH_CONTROL, ODM, inst), \ SRI_ARR(OPTC_WIDTH_CONTROL2, ODM, inst), \ SRI_ARR(OPTC_MEMORY_CONFIG, ODM, inst), \ - SRI_ARR(OTG_DRR_CONTROL, OTG, inst), \ - SRI_ARR(OTG_PSTATE_REGISTER, OTG, inst) + SRI_ARR(OTG_DRR_CONTROL, OTG, inst), \ + SRI_ARR(OTG_PSTATE_REGISTER, OTG, inst), \ + SRI_ARR(OTG_PIPE_UPDATE_STATUS, OTG, inst) /* HUBBUB */ #define HUBBUB_REG_LIST_DCN4_01_RI(id) \ -- 2.51.0 From 3fc98044eb33f1ed11cc708253b115e4325cc167 Mon Sep 17 00:00:00 2001 From: Zhongwei Date: Tue, 10 Sep 2024 13:28:34 +0800 Subject: [PATCH 10/16] drm/amd/display: Monitor patch to call blank_stream() before otg off [Why] Turning off OTG before DIG is on and backlight is on, might cause: DIG FIFO underflow. EDP output unexpected video data. That might violate EDP spec. EDP spec requires black light should be off before validate video data is turned off. Currently garbage issue only occurs in one type of EDP to MIPI converter. [How] Add monitor patch to call blank_stream() before disable_crtc(). That will be no impact for current sequence. If there are more types of EDP panels meeting this issue later, we might remove this monitor patch and fix the current sequence. Reviewed-by: Nicholas Kazlauskas Signed-off-by: Zhongwei Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc_types.h | 1 + drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c | 5 +++++ 2 files changed, 6 insertions(+) diff --git a/drivers/gpu/drm/amd/display/dc/dc_types.h b/drivers/gpu/drm/amd/display/dc/dc_types.h index fd6dca735714..2bbafd1cdce4 100644 --- a/drivers/gpu/drm/amd/display/dc/dc_types.h +++ b/drivers/gpu/drm/amd/display/dc/dc_types.h @@ -178,6 +178,7 @@ struct dc_panel_patch { unsigned int skip_avmute; unsigned int mst_start_top_delay; unsigned int remove_sink_ext_caps; + uint8_t blankstream_before_otg_off; }; struct dc_edid_caps { diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c index 3d4b31bd9946..bfc78a42bc2a 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn31/dcn31_hwseq.c @@ -517,6 +517,11 @@ static void dcn31_reset_back_end_for_pipe( dc->hwss.set_abm_immediate_disable(pipe_ctx); + if ((!pipe_ctx->stream->dpms_off || pipe_ctx->stream->link->link_status.link_active) + && pipe_ctx->stream->sink && pipe_ctx->stream->sink->edid_caps.panel_patch.blankstream_before_otg_off) { + dc->hwss.blank_stream(pipe_ctx); + } + pipe_ctx->stream_res.tg->funcs->set_dsc_config( pipe_ctx->stream_res.tg, OPTC_DSC_DISABLED, 0, 0); -- 2.51.0 From 4641169a8c95d9efc35d2d3c55c3948f3b375ff9 Mon Sep 17 00:00:00 2001 From: Fangzhi Zuo Date: Wed, 4 Sep 2024 16:56:45 -0400 Subject: [PATCH 11/16] drm/amd/display: Fix incorrect DSC recompute trigger A stream without dsc_aux should not be eliminated from the dsc determination. Whether it needs a dsc recompute depends on whether its mode has changed or not. Eliminating such a no-dsc stream from the dsc determination policy will end up with inconsistencies in the new dc_state when compared to the current dc_state, triggering a dsc recompute that should not have happened. Reviewed-by: Rodrigo Siqueira Signed-off-by: Fangzhi Zuo Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c index a08e8a0b696c..68bd8e6b948d 100644 --- a/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c +++ b/drivers/gpu/drm/amd/display/amdgpu_dm/amdgpu_dm_mst_types.c @@ -1302,7 +1302,7 @@ static bool is_dsc_need_re_compute( continue; aconnector = (struct amdgpu_dm_connector *) stream->dm_stream_context; - if (!aconnector || !aconnector->dsc_aux) + if (!aconnector) continue; stream_on_link[new_stream_on_link_num] = aconnector; -- 2.51.0 From 3924f200f199e604bc3a9f965b6dd7617d2bbe6a Mon Sep 17 00:00:00 2001 From: Sung Lee Date: Mon, 9 Sep 2024 16:12:55 -0400 Subject: [PATCH 12/16] drm/amd/display: Clean up triple buffer enablement code [WHY] Triple buffer enablement currently does not work properly [HOW] Allow triple buffer enablement to happen properly on fast updates Reviewed-by: Aric Cyr Signed-off-by: Sung Lee Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 5c39390ecbd5..9d427a3d1efb 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -3677,13 +3677,14 @@ static void commit_planes_for_stream_fast(struct dc *dc, if (!pipe_ctx->plane_state) continue; - if (should_update_pipe_for_plane(context, pipe_ctx, plane_state)) + if (!should_update_pipe_for_plane(context, pipe_ctx, plane_state)) continue; + pipe_ctx->plane_state->triplebuffer_flips = false; if (update_type == UPDATE_TYPE_FAST && - dc->hwss.program_triplebuffer != NULL && - !pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) { - /*triple buffer for VUpdate only*/ + dc->hwss.program_triplebuffer != NULL && + !pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) { + /*triple buffer for VUpdate only*/ pipe_ctx->plane_state->triplebuffer_flips = true; } } @@ -3920,19 +3921,20 @@ static void commit_planes_for_stream(struct dc *dc, struct pipe_ctx *pipe_ctx = &context->res_ctx.pipe_ctx[j]; if (!pipe_ctx->plane_state) continue; - if (should_update_pipe_for_plane(context, pipe_ctx, plane_state)) + if (!should_update_pipe_for_plane(context, pipe_ctx, plane_state)) continue; pipe_ctx->plane_state->triplebuffer_flips = false; if (update_type == UPDATE_TYPE_FAST && - dc->hwss.program_triplebuffer != NULL && - !pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) { - /*triple buffer for VUpdate only*/ - pipe_ctx->plane_state->triplebuffer_flips = true; + dc->hwss.program_triplebuffer != NULL && + !pipe_ctx->plane_state->flip_immediate && dc->debug.enable_tri_buf) { + /*triple buffer for VUpdate only*/ + pipe_ctx->plane_state->triplebuffer_flips = true; } } if (update_type == UPDATE_TYPE_FULL) { /* force vsync flip when reconfiguring pipes to prevent underflow */ plane_state->flip_immediate = false; + plane_state->triplebuffer_flips = false; } } @@ -3953,7 +3955,6 @@ static void commit_planes_for_stream(struct dc *dc, continue; ASSERT(!pipe_ctx->plane_state->triplebuffer_flips); - if (dc->hwss.program_triplebuffer != NULL && dc->debug.enable_tri_buf) { /*turn off triple buffer for full update*/ dc->hwss.program_triplebuffer( @@ -4028,7 +4029,7 @@ static void commit_planes_for_stream(struct dc *dc, /*program triple buffer after lock based on flip type*/ if (dc->hwss.program_triplebuffer != NULL && dc->debug.enable_tri_buf) { - /*only enable triplebuffer for fast_update*/ + /*only enable triplebuffer for fast_update*/ dc->hwss.program_triplebuffer( dc, pipe_ctx, pipe_ctx->plane_state->triplebuffer_flips); } -- 2.51.0 From 83762925c94c9053de9aedf0a880907a655ec394 Mon Sep 17 00:00:00 2001 From: Shunlu Zhang Date: Tue, 3 Sep 2024 01:08:01 -0400 Subject: [PATCH 13/16] drm/amd/display: Change dc_surface_update cm2_params to const [WHY] cm2_params is used to set update_flags. It's value is not intended to be modified. [WHAT] Change the declaration of cm2_params to be a constant variable Reviewed-by: Tao Huang Reviewed-by: Ariel Bernstein Signed-off-by: Shunlu Zhang Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/dc.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 3992ad73165b..7b164e8fa104 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -1513,7 +1513,7 @@ struct dc_surface_update { * change cm2_params.component_settings: Full update * change cm2_params.cm2_luts: Fast update */ - struct dc_cm2_parameters *cm2_params; + const struct dc_cm2_parameters *cm2_params; const struct dc_csc_transform *cursor_csc_color_matrix; unsigned int sdr_white_level_nits; }; -- 2.51.0 From 0e719a40bf5234e66bbd0f1537e30709ab112b5f Mon Sep 17 00:00:00 2001 From: Austin Zheng Date: Tue, 10 Sep 2024 13:46:50 -0400 Subject: [PATCH 14/16] drm/amd/display: Set Pipe Unlock Order Outside of HWSEQ [Why] Current pipe unlock order set within dcn401_interdependent_update_lock. Separate the logic from the functionality to allow for unit testing and make it easier to debug. [How] Add a flag to indicate if a pipe unlock order should be set. Create function to determine unlock order. Indicate which pipes should be unlocked first using array stored in dc scratch memory. Pipes indicated in array can be unlocked in any order. Reviewed-by: Alvin Lee Signed-off-by: Austin Zheng Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/core/dc.c | 42 +++++++++++++++++++ drivers/gpu/drm/amd/display/dc/dc.h | 2 + .../amd/display/dc/hwss/dcn401/dcn401_hwseq.c | 32 ++++++-------- .../dc/resource/dcn401/dcn401_resource.c | 1 + 4 files changed, 57 insertions(+), 20 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/core/dc.c b/drivers/gpu/drm/amd/display/dc/core/dc.c index 9d427a3d1efb..f87f12a7be90 100644 --- a/drivers/gpu/drm/amd/display/dc/core/dc.c +++ b/drivers/gpu/drm/amd/display/dc/core/dc.c @@ -1876,6 +1876,41 @@ void dc_z10_save_init(struct dc *dc) dc->hwss.z10_save_init(dc); } +/* Set a pipe unlock order based on the change in DET allocation and stores it in dc scratch memory + * Prevents over allocation of DET during unlock process + * e.g. 2 pipe config with different streams with a max of 20 DET segments + * Before: After: + * - Pipe0: 10 DET segments - Pipe0: 12 DET segments + * - Pipe1: 10 DET segments - Pipe1: 8 DET segments + * If Pipe0 gets updated first, 22 DET segments will be allocated + */ +static void determine_pipe_unlock_order(struct dc *dc, struct dc_state *context) +{ + unsigned int i = 0; + struct pipe_ctx *pipe = NULL; + struct timing_generator *tg = NULL; + + if (!dc->config.set_pipe_unlock_order) + return; + + memset(dc->scratch.pipes_to_unlock_first, 0, sizeof(dc->scratch.pipes_to_unlock_first)); + for (i = 0; i < dc->res_pool->pipe_count; i++) { + pipe = &context->res_ctx.pipe_ctx[i]; + tg = pipe->stream_res.tg; + + if (!resource_is_pipe_type(pipe, OTG_MASTER) || + !tg->funcs->is_tg_enabled(tg) || + dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) { + continue; + } + + if (resource_calculate_det_for_stream(context, pipe) < + resource_calculate_det_for_stream(dc->current_state, &dc->current_state->res_ctx.pipe_ctx[i])) { + dc->scratch.pipes_to_unlock_first[i] = true; + } + } +} + /** * dc_commit_state_no_check - Apply context to the hardware * @@ -1974,6 +2009,7 @@ static enum dc_status dc_commit_state_no_check(struct dc *dc, struct dc_state *c context->streams[i]->update_flags.bits.dsc_changed = prev_dsc_changed; } + determine_pipe_unlock_order(dc, context); /* Program all planes within new context*/ if (dc->res_pool->funcs->prepare_mcache_programming) dc->res_pool->funcs->prepare_mcache_programming(dc, context); @@ -3625,6 +3661,10 @@ static void commit_planes_for_stream_fast(struct dc *dc, struct pipe_ctx *top_pipe_to_program = NULL; struct dc_stream_status *stream_status = NULL; bool should_offload_fams2_flip = false; + bool should_lock_all_pipes = (update_type != UPDATE_TYPE_FAST); + + if (should_lock_all_pipes) + determine_pipe_unlock_order(dc, context); if (dc->debug.fams2_config.bits.enable && dc->debug.fams2_config.bits.enable_offload_flip && @@ -3743,6 +3783,8 @@ static void commit_planes_for_stream(struct dc *dc, bool subvp_curr_use = false; uint8_t current_stream_mask = 0; + if (should_lock_all_pipes) + determine_pipe_unlock_order(dc, context); // Once we apply the new subvp context to hardware it won't be in the // dc->current_state anymore, so we have to cache it before we apply // the new SubVP context diff --git a/drivers/gpu/drm/amd/display/dc/dc.h b/drivers/gpu/drm/amd/display/dc/dc.h index 7b164e8fa104..d3b6a389fece 100644 --- a/drivers/gpu/drm/amd/display/dc/dc.h +++ b/drivers/gpu/drm/amd/display/dc/dc.h @@ -463,6 +463,7 @@ struct dc_config { unsigned int enable_fpo_flicker_detection; bool disable_hbr_audio_dp2; bool consolidated_dpia_dp_lt; + bool set_pipe_unlock_order; }; enum visual_confirm { @@ -1461,6 +1462,7 @@ struct dc { struct dc_scratch_space current_state; struct dc_scratch_space new_state; struct dc_stream_state temp_stream; // Used so we don't need to allocate stream on the stack + bool pipes_to_unlock_first[MAX_PIPES]; /* Any of the pipes indicated here should be unlocked first */ } scratch; struct dml2_configuration_options dml2_options; diff --git a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c index 0b743669f23b..29fcb4ed6646 100644 --- a/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c +++ b/drivers/gpu/drm/amd/display/dc/hwss/dcn401/dcn401_hwseq.c @@ -1705,7 +1705,6 @@ void dcn401_interdependent_update_lock(struct dc *dc, unsigned int i = 0; struct pipe_ctx *pipe = NULL; struct timing_generator *tg = NULL; - bool pipe_unlocked[MAX_PIPES] = {0}; if (lock) { for (i = 0; i < dc->res_pool->pipe_count; i++) { @@ -1719,43 +1718,36 @@ void dcn401_interdependent_update_lock(struct dc *dc, dc->hwss.pipe_control_lock(dc, pipe, true); } } else { - /* Unlock pipes based on the change in DET allocation instead of pipe index - * Prevents over allocation of DET during unlock process - * e.g. 2 pipe config with different streams with a max of 20 DET segments - * Before: After: - * - Pipe0: 10 DET segments - Pipe0: 12 DET segments - * - Pipe1: 10 DET segments - Pipe1: 8 DET segments - * If Pipe0 gets updated first, 22 DET segments will be allocated - */ + /* Need to free DET being used first and have pipe update, then unlock the remaining pipes*/ for (i = 0; i < dc->res_pool->pipe_count; i++) { pipe = &context->res_ctx.pipe_ctx[i]; tg = pipe->stream_res.tg; - int current_pipe_idx = i; if (!resource_is_pipe_type(pipe, OTG_MASTER) || !tg->funcs->is_tg_enabled(tg) || dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) { - pipe_unlocked[i] = true; continue; } - // If the same stream exists in old context, ensure the OTG_MASTER pipes for the same stream get compared - struct pipe_ctx *old_otg_master = resource_get_otg_master_for_stream(&dc->current_state->res_ctx, pipe->stream); - - if (old_otg_master) - current_pipe_idx = old_otg_master->pipe_idx; - if (resource_calculate_det_for_stream(context, pipe) < - resource_calculate_det_for_stream(dc->current_state, &dc->current_state->res_ctx.pipe_ctx[current_pipe_idx])) { + if (dc->scratch.pipes_to_unlock_first[i]) { dc->hwss.pipe_control_lock(dc, pipe, false); - pipe_unlocked[i] = true; dcn401_wait_for_det_buffer_update(dc, context, pipe); } } + /* Unlocking the rest of the pipes */ for (i = 0; i < dc->res_pool->pipe_count; i++) { - if (pipe_unlocked[i]) + if (dc->scratch.pipes_to_unlock_first[i]) continue; + pipe = &context->res_ctx.pipe_ctx[i]; + tg = pipe->stream_res.tg; + if (!resource_is_pipe_type(pipe, OTG_MASTER) || + !tg->funcs->is_tg_enabled(tg) || + dc_state_get_pipe_subvp_type(context, pipe) == SUBVP_PHANTOM) { + continue; + } + dc->hwss.pipe_control_lock(dc, pipe, false); } } diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c index 9d56fbdcd06a..f2653a86d3e7 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn401/dcn401_resource.c @@ -1867,6 +1867,7 @@ static bool dcn401_resource_construct( dc->config.prefer_easf = true; dc->config.dc_mode_clk_limit_support = true; dc->config.enable_windowed_mpo_odm = true; + dc->config.set_pipe_unlock_order = true; /* Need to ensure DET gets freed before allocating */ /* read VBIOS LTTPR caps */ { if (ctx->dc_bios->funcs->get_lttpr_caps) { -- 2.51.0 From 4f01a68751194d05280d659a65758c09e4af04d6 Mon Sep 17 00:00:00 2001 From: Michael Strauss Date: Tue, 3 Sep 2024 16:23:11 -0400 Subject: [PATCH 15/16] drm/amd/display: Block UHBR Based On USB-C PD Cable ID [WHY] Currently the absence of UHBR cable ID caps from USB-C PD does not block UHBR rates. In situations where DPCD reports valid UHBR capability but USB-C PD does not, such as using a USB-C to DP1.4 dongle connected to a native DP2.1 cable, link loss and lightup failures can be seen as a result. Additionally, in edge cases where a platform supports cable ID but DMUB doesn't correctly return cable ID caps, driver currently also allows UHBR. [HOW] Block UHBR rates over DP alt mode if cable id indicates no UHBR support. Additionally, block UHBR rates if a cable ID supported platform receives no reply from a DMUB cable id query. Reviewed-by: Wenjing Liu Reviewed-by: Ovidiu Bunea Signed-off-by: Michael Strauss Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- .../dc/link/protocols/link_dp_capability.c | 22 ++++++++++++++----- 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c index d78c8ec4de79..fe4aa2c158ea 100644 --- a/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c +++ b/drivers/gpu/drm/amd/display/dc/link/protocols/link_dp_capability.c @@ -1409,7 +1409,8 @@ static bool get_usbc_cable_id(struct dc_link *link, union dp_cable_id *cable_id) if (!link->ctx->dmub_srv || link->ep_type != DISPLAY_ENDPOINT_PHY || - link->link_enc->features.flags.bits.DP_IS_USB_C == 0) + link->link_enc->features.flags.bits.DP_IS_USB_C == 0 || + link->link_enc->features.flags.bits.IS_DP2_CAPABLE == 0) return false; memset(&cmd, 0, sizeof(cmd)); @@ -1422,7 +1423,9 @@ static bool get_usbc_cable_id(struct dc_link *link, union dp_cable_id *cable_id) cable_id->raw = cmd.cable_id.data.output_raw; DC_LOG_DC("usbc_cable_id = %d.\n", cable_id->raw); } - return cmd.cable_id.header.ret_status == 1; + + ASSERT(cmd.cable_id.header.ret_status); + return true; } static void retrieve_cable_id(struct dc_link *link) @@ -2103,6 +2106,8 @@ struct dc_link_settings dp_get_max_link_cap(struct dc_link *link) /* get max link encoder capability */ if (link_enc) link_enc->funcs->get_max_link_cap(link_enc, &max_link_cap); + else + return max_link_cap; /* Lower link settings based on sink's link cap */ if (link->reported_link_cap.lane_count < max_link_cap.lane_count) @@ -2136,10 +2141,15 @@ struct dc_link_settings dp_get_max_link_cap(struct dc_link *link) */ cable_max_link_rate = get_cable_max_link_rate(link); - if (!link->dc->debug.ignore_cable_id && - cable_max_link_rate != LINK_RATE_UNKNOWN) { - if (cable_max_link_rate < max_link_cap.link_rate) - max_link_cap.link_rate = cable_max_link_rate; + if (!link->dc->debug.ignore_cable_id) { + if (cable_max_link_rate != LINK_RATE_UNKNOWN) + // cable max link rate known + max_link_cap.link_rate = MIN(max_link_cap.link_rate, cable_max_link_rate); + else if (link_enc->funcs->is_in_alt_mode && link_enc->funcs->is_in_alt_mode(link_enc)) + // cable max link rate ambiguous, DP alt mode, limit to HBR3 + max_link_cap.link_rate = MIN(max_link_cap.link_rate, LINK_RATE_HIGH3); + //else {} + // cable max link rate ambiguous, DP, do nothing if (!link->dpcd_caps.cable_id.bits.UHBR13_5_CAPABILITY && link->dpcd_caps.cable_id.bits.CABLE_TYPE >= 2) -- 2.51.0 From 1c6b16ebf5eb2bc5740be9e37b3a69f1dfe1dded Mon Sep 17 00:00:00 2001 From: Charlene Liu Date: Wed, 11 Sep 2024 19:45:09 -0400 Subject: [PATCH 16/16] drm/amd/display: avoid set dispclk to 0 [why] set dispclk to 0 cause stability issue. Cc: Mario Limonciello Cc: Alex Deucher Reviewed-by: Nicholas Kazlauskas Signed-off-by: Charlene Liu Signed-off-by: Aurabindo Pillai Tested-by: Daniel Wheeler Signed-off-by: Alex Deucher --- drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c | 1 + 1 file changed, 1 insertion(+) diff --git a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c index da9101b83e8c..70abd32ce2ad 100644 --- a/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c +++ b/drivers/gpu/drm/amd/display/dc/resource/dcn351/dcn351_resource.c @@ -766,6 +766,7 @@ static const struct dc_debug_options debug_defaults_drv = { .disable_dmub_reallow_idle = false, .static_screen_wait_frames = 2, .notify_dpia_hr_bw = true, + .min_disp_clk_khz = 50000, }; static const struct dc_panel_config panel_config_defaults = { -- 2.51.0